def __call__(self, data):
"""
Mapper Program
"""
true_cnt_t = 0
true_cnt_f = 0
false_cnt_t = 0
false_cnt_f = 0
for docID, doc in data:
for term in doc.split("\n"):
self.SEP = self.SEP if self.SEP is not None else get_sep(term)
point = np.fromstring(term, dtype=np.float64, sep=self.SEP)
result = self.varify(point)
if result == TRUE_F:
true_cnt_f = true_cnt_f + 1
elif result == TRUE_T:
true_cnt_t = true_cnt_t + 1
elif result == FALSE_F:
false_cnt_f = false_cnt_f + 1
elif result == FALSE_T:
false_cnt_t = false_cnt_t + 1
yield TRUE_T_STR, true_cnt_t
yield TRUE_F_STR, true_cnt_f
yield FALSE_T_STR, false_cnt_t
yield FALSE_F_STR, false_cnt_f
def __call__(self, data):
"""
Mapper Program
"""
true_cnt_t = 0
true_cnt_f = 0
false_cnt_t = 0
false_cnt_f = 0
for docID, doc in data:
for term in doc.split("\n"):
self.SEP = self.SEP if self.SEP is not None else get_sep(term)
point = np.fromstring(term, dtype=np.float64, sep=self.SEP)
result = self.varify(point)
if result == TRUE_F:
true_cnt_f = true_cnt_f + 1
elif result == TRUE_T:
true_cnt_t = true_cnt_t + 1
elif result == FALSE_F:
false_cnt_f = false_cnt_f + 1
elif result == FALSE_T:
false_cnt_t = false_cnt_t + 1
yield TRUE_T_STR, true_cnt_t
yield TRUE_F_STR, true_cnt_f
yield FALSE_T_STR, false_cnt_t
yield FALSE_F_STR, false_cnt_f
def __call__(self, data):
"""
Mapper Program:
Inputs:
data, which is the whole split block data
Outputs:
key: untified id
value: resultD,resultH
"""
# SETP1: read data matrix and do some transpose
resultH = None
resultD = None
for docID, doc in data:
for term in doc.split("\n"):
self.SEP = self.SEP if self.SEP is not None else get_sep(term)
point = np.fromstring(term, dtype=np.float64, sep=self.SEP)
(localH, localD) = self.calculate(point)
if resultH is not None:
resultH = resultH + localH
resultD = resultD + localD
else:
resultH = localH
resultD = localD
debug(np.shape(resultH))
debug(np.shape(resultD))
yield "nonused", (resultD.tolist(), resultH.tolist())
def __call__(self, data):
"""
Mapper Program:
Inputs:
data, which is the whole split block data
Outputs:
key: untified id
value: resultD,resultH
"""
# SETP1: read data matrix and do some transpose
resultH = None
resultD = None
for docID, doc in data:
for term in doc.split("\n"):
self.SEP = self.SEP if self.SEP is not None else get_sep(term)
point = np.fromstring(term, dtype=np.float64, sep=self.SEP)
(localH, localD) = self.calculate(point)
if resultH is not None:
resultH = resultH + localH
resultD = resultD + localD
else:
resultH = localH
resultD = localD
debug(np.shape(resultH))
debug(np.shape(resultD))
yield "nonused", (resultD.tolist(), resultH.tolist())
def __call__(self, data):
"""
Mapper Program
"""
cnt = 0
means = 0.0
for docID, doc in data:
for term in doc.split("\n"):
self.SEP = self.SEP if self.SEP is not None else get_sep(term)
point = np.fromstring(term, dtype=np.float64, sep=self.SEP)
means += self.getDValue(point)
cnt = cnt + 1
yield "nonused", (cnt, means)
def __call__(self, data):
"""
Mapper Program
It will output the modified single line
"""
for docID, doc in data:
for term in doc.split("\n"):
self.SEP = self.SEP if self.SEP is not None else get_sep(term)
point = np.fromstring(term, dtype=np.float64, sep=self.SEP)
label = int(point[-1])
last_value = self.getDValue(point)
point = self.extend_point(point)
point[-1] = last_value
point[-2] = float(label)
output = ",".join([str(i) for i in point])
yield output, "\t"